Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Evaluating Moral Beliefs across LLMs through a Pluralistic Framework

Xuelin Liu, Yanfei Zhu, Shucheng Zhu, Pengyuan Liu, Ying Liu, Dong Yu

TL;DR

A novel three-module framework to evaluate the moral beliefs of four prominent large language models is introduced, indicating that English language models, namely ChatGPT and Gemini, closely mirror moral decisions of the sample of Chinese university students, demonstrating strong adherence to their choices and a preference for individualistic moral beliefs.

Abstract

Proper moral beliefs are fundamental for language models, yet assessing these beliefs poses a significant challenge. This study introduces a novel three-module framework to evaluate the moral beliefs of four prominent large language models. Initially, we constructed a dataset containing 472 moral choice scenarios in Chinese, derived from moral words. The decision-making process of the models in these scenarios reveals their moral principle preferences. By ranking these moral choices, we discern the varying moral beliefs held by different language models. Additionally, through moral debates, we investigate the firmness of these models to their moral choices. Our findings indicate that English language models, namely ChatGPT and Gemini, closely mirror moral decisions of the sample of Chinese university students, demonstrating strong adherence to their choices and a preference for individualistic moral beliefs. In contrast, Chinese models such as Ernie and ChatGLM lean towards collectivist moral beliefs, exhibiting ambiguity in their moral choices and debates. This study also uncovers gender bias embedded within the moral beliefs of all examined language models. Our methodology offers an innovative means to assess moral beliefs in both artificial and human intelligence, facilitating a comparison of moral values across different cultures.

Evaluating Moral Beliefs across LLMs through a Pluralistic Framework

TL;DR

A novel three-module framework to evaluate the moral beliefs of four prominent large language models is introduced, indicating that English language models, namely ChatGPT and Gemini, closely mirror moral decisions of the sample of Chinese university students, demonstrating strong adherence to their choices and a preference for individualistic moral beliefs.

Abstract

Proper moral beliefs are fundamental for language models, yet assessing these beliefs poses a significant challenge. This study introduces a novel three-module framework to evaluate the moral beliefs of four prominent large language models. Initially, we constructed a dataset containing 472 moral choice scenarios in Chinese, derived from moral words. The decision-making process of the models in these scenarios reveals their moral principle preferences. By ranking these moral choices, we discern the varying moral beliefs held by different language models. Additionally, through moral debates, we investigate the firmness of these models to their moral choices. Our findings indicate that English language models, namely ChatGPT and Gemini, closely mirror moral decisions of the sample of Chinese university students, demonstrating strong adherence to their choices and a preference for individualistic moral beliefs. In contrast, Chinese models such as Ernie and ChatGLM lean towards collectivist moral beliefs, exhibiting ambiguity in their moral choices and debates. This study also uncovers gender bias embedded within the moral beliefs of all examined language models. Our methodology offers an innovative means to assess moral beliefs in both artificial and human intelligence, facilitating a comparison of moral values across different cultures.

Paper Structure

This paper contains 40 sections, 4 figures, 13 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Moral theories
  4. Moral beliefs in LLMs
  5. Debate
  6. Dataset
  7. Moral word list
  8. Moral scenario
  9. Experiment
  10. Experimental setups
  11. Moral judgments based on moral word
  12. Moral choice based on moral scenario
  13. Moral rank
  14. Moral debate
  15. Results
  16. ...and 25 more sections

Figures (4)

  • Figure 1: Our three-module framework to evaluate LLMs' moral beliefs, including moral choice, moral rank, and moral debate.
  • Figure 2: The consistency of moral choice among 4 models and the sample of Chinese university students (SCU). The higher the number, the darker the color, indicating a higher similarity.
  • Figure 3: Correlation between the ranks of 4 models and the sample of Chinese university students (SCU). We use the Spearman’s rank-order correlation coefficient (SROCC) to measure the correlation. The darker the color, the stronger the correlation.
  • Figure 4: The proportion of changes in model options before and after debate. The darker the color, the more indecisive the model tends to be during debates, making it more prone to changing its choices.